1—Solid Phase Chemical Libraries
An emerging paradigm for lead discovery in pharmaceutical and related industries such as agricultural biotechnology, is the assembly of novel synthetic compound libraries by new methods of solid state “combinatorial” synthesis. Combinatorial chemistry refers to a set of strategies for the parallel synthesis and testing of multiple compounds or compounds mixtures, either in solution or in solid supports in the form of beaded resins (“beads”). In general, a combinatorial synthesis employing M precursors in each of N reaction steps produces M^N compounds. For example, a combinatorial synthesis produces 4^N oligonnucleotides in N steps, each employing 4 oligonucleotide precursors; similarly, a combinatorial synthesis of N steps, each employing 20 amino acid precursors, produces 20^N oligopeptides.
1.1—One Bead/One Compound Chemical Libraries
One implementation of combinatorial synthesis that is suitable to produce very large chemical libraries relies on solid supports in the form of beaded resins (“beads”) and encodes reaction steps in a “divide, couple and recombine” (DCR) strategy (FIG. 1), also refereed to as “resin-splitting” synthesis. The resulting “one bead/one compound” chemical libraries contain from 10^6 to 10^8 compounds. These libraries are screened by performing a wide variety of chemical and biochemical assays to identify individual compounds eliciting a positive response. The chemical identity of such compounds can be determined by direct analysis.
Two methods of direct analysis are micro-sequencing and mass spectrometry. Both methods require the physical isolation of synthesis beads displaying compounds of interest and both require off-line chemical analysis based on substantial amounts of compound—tens to hundreds of picomoles. Micro-sequencing, limited to libraries of oligopeptides and oligonucleotides, does not distinguish between stereoisomers. Mass spectrometry is unable to distinguish between precursors of equal mass such as D-and L-amino acids or leucine and isoleucine. The requirement of direct chemical analysis for a substantial quantity of compound dictates the use of large bead resins (a typical bead diameter is 130 μm) to ensure that picomolar quantities of each compound can be recovered, even when it is becoming increasingly desirable to perform high throughput screening of the compound library in miniaturized environments to reduce requisite volumes of sample and reagents and to enhance throughput.
1.2 —Encoded One Bead/One Component Chemical Libraries
One approach to overcoming the serious limitations of standard one bead/one compound chemical libraries is to encode chemical compound identities. This facilitates the identification of compounds not amenable to direct determination by micro-sequencing or mass spectrometry. One encoding method employs the co-synthesis of peptides and oligonucleotides to represent the identity of non-sequenceable synthesis products (Nikolaiev et al., “Peptide-Encoding for Structure Determination of Non-Sequenceable Polymers Within Libraries Synthesized and Tested on Solid-Phase Supports”, Peptides Res. 6, 161 (1993), the contents of which are included herein by reference). A second method, compatible with a wider range of chemical reaction conditions, employs a set of tagging molecules to record the reaction histories of beads.
One implementation of the latter method uses a set of pre-synthesized, chromatographically distinguishable molecular tags T1, T2, . . . , TM to construct a chemical binary code. In prior art, molecular tags are structurally related molecules (FIG. 2) which can be identified by their characteristic gas chromatographic retention times (Still et al., “Complex combinatorial libraries encoded with tags”, U.S. Pat. No. 5,565,324, the contents of which are included herein by reference).
At each step of DCR synthesis, a unique tag from the set is added to each divided aliquot to record the reaction carried out with that aliquot. The concept may be illustrated by examining the steps of a 2-step synthesis using reagents R11, R21 and R31 in step 1, and reagents R21, R22 and R23 in step 2, to generate nine products. The reagents of the first step are uniquely identified by the binary addresses 01 (R11) 10(R12) and 11(R13), and the reagents of the second step are uniquely identified by the binary addresses 01 (R21), 10(R22) and 11 (R23Each binary address is chemically represented in terms of a set of molecular tags: T1 (01 in step 1 representing R11), T2 (10 in step 1 representing R12) and T2T1 (11 in step 1 representing R13) and analogously with T3 (01 in step 2 representing R21), T4 (10 in step 2 repre senting R22) and T4T3 (11 in step 2 representing R23).
A sequence of reaction steps is recorded by simply concatenating binary addresses. Thus, 11.01, read right to left, would indicate the sequence “reagent R23 in step 2, reagent R11 in step 1”. The chemical representation of this sequence is T4T3. T1, and the presence on the bead of this particular set of tags indicates the chemical identity of the bead-anchored synthesis product. The strategy is readily generalized to larger reactions. For example, 7 reagents to be used in each reaction step can be uniquely identified by the binary addresses 001 (R11), 010 (R12), . . . , 111 (R17). Although superior to un-encoded one bead/one compound methods, nevertheless the tagging strategy of prior art still suffer from three limitations. First, individual beads of interest must be physically isolated from the rest; next, molecular tags must be chemically or photochemically cleaved from the bead and cleaved tags must be collected; and finally, chemical analysis (e.g., gas chromatography) must be performed. These numerous time-and labor-intensive manipulations eliminate much of the enhancement in throughput gained by the DCR synthesis strategy.
1.3 Screening and Lead Compound Optimization
The high specificity of typical biological substrate-target interactions implies that the vast majority of compounds in a library will be inactive for any particular target. Thus, the task of screening is to identify the very few compounds within the library that display activity in binding or in functional assays. Common targets include enzymes and receptors as well as nucleic acids.
To implement the rapid screening and scoring of an entire library of synthetic compounds, in practice containing 10^4 to 10^8 compounds, requires systematic screening procedures if the task is to be completed within viable time frames. Several assay formats have been described to implement the screening of bead-based combinatorial libraries. These include: reaction of a collection of beads, allowed to settle under gravity, with an enzyme-labeled or fluorophore-labeled target molecule followed by visual detection (Lam et al., “A new type of synthetic peptide library for identifying ligand-binding activity”, Nature 354 (1991), the contents of which are included herein by reference); incubation of beads with radio-labeled target molecules and subsequent agarose immobilization of beads and auto-radiographic detection (Kassarjian, Schellenberger and Turck, “Screening of Synthetic Peptide Libraries with Radio-labeled Acceptor Molecules”, Peptide Res. 6, 129 (1993), the contents of which are included herein by reference); and partial release of compounds from beads for solution-phase testing (Salmon et al., “Discovery of biologically active peptides in random libraries: Solution-phase testing after staged orthogonal release from resin beads”, Proc. Natl. Acad. Sc. USA 90, 11708 (1993), the contents of which are included herein by reference).
WO95/32425 provides a method of preparing combinational libraries using a method of encoding combinational libraries with fluorophore labeled beads. According to the method, a first combinational library is prepared by conducting a set of reactions on tagged beads to afford an encoded first registry (i.e., step in the synthetic sequence). A second combinational library is prepared using similar reaction steps but the tagged beads are combined and separated prior to the first reaction sequence and the beads are sorted prior to the second reaction sequence. Subsequent libraries are prepared as for the second library except that the sorting step takes place prior to a different registry in each subsequent library. Thus, WO95/32425 teaches only individually labelling the first step and physical separatois of beads to identify each modified combinational library.
Nederlof et al., Cytometry, 13, 839–845 (1992), teaches the use of ratio labeling as a way of increasing the number of simultaneously detectable probes beyond the seven used previously. In this approach, ratio-labelled probes are identified on the basis of the ratio of color intensity, not just the particular colors used. Fluorescence ratios are measured and used as additional encoding colors. The method requires double-labeling of probes using different ratios of labels. The method is not specifically directed to synthetic combinational libraries. Accordingly, the field of Nederlof's method is the detection of multiple DNA/RNA sequence by in situ hybridization, and is not relevant to the field of encoding of synthetic chemical libraries.
Speiche, Ballard & Ward, Nature Genetics, 12, 368 (1996), describe a method of characterizing complex chromosomal karyo types using multi-fluorescence in situ hybridization. Instead of using ratio-double labelling as in Nederlof, Speiche et al. use a set of six fluorescent dyes with spectral emission peaks spread across the photometric response range to visualize 27 combinationally labelled probes. Speiche et al. do not disclose a method of encoding synthetic combinational libraries.
Still et al., Proc. Nat'l Acad. Sci., 90, 10922–926 (1993), disclose a method of synthesis of tagged combinational libraries using a binary code based on different electrophoric tags. The method requires use of photocleavable molecular tags which comprise variously substituted aryl moieties linked via a variable-length aliphatic hydrocarbon chain, whereby the tags when cleaved are distinctly resolvable by capillary gas chromatography with electochemical detection. Color detection is not used in this method. The method also requires cleavage from the solid support in order to analyze the sequence. In related work, Still et al. U.S. Pat. No. 5,721,099 disclose methods of preparing encoded combinatorial libraries, but again the method requires cleavage of the identifier tags prior to analysis of the encoded reaction history. In contrast, the present invention provides an in situ approach to the interrogation of encoded combinatorial libraries, and represents an advance over the prior methods of encoding libraries. The success of the present invention is unexpected in view of the prior approaches because of the scattering phenomena expected for a spectral analysis performed in heterogeneous media which would dissipate spectral signal-to-noise giving rise to practical difficulties in detecting accurately relative abundance information for fluorophore tags. The present methodology demonstrates for the first time a way of solving these practical problems in performing in situ encoding and interrogation of combinatorial libraries.
II —Multi-Agent Monitoring and Diagnostics
Diagnostic panels display multiple chemistries to screen unknown solutions for the presence of multiple agents. For example, blood group specificity is determined by spotting an unknown blood sample onto a panel of surface-bound antibodies whose arrangement in the panel reflects their antigen-specificity. Antigen-binding to any specific patch in the panel reveals the chemical identify of the antigen and enhance the blood type. Another realization of the same concept of displaying multiple diagnostic probes in a spatially encoded panel or array involves screening of mutations by assaying for hybridization of DNA to one of a large number of candidate matching strands which are placed in known positions on a planar substrate in a checkerboard pattern. This may be achieved by dispensing droplets containing distinct probes, or may involve the in-situ synthesis of oligonucleotide strands of varying composition.
Spatial encoding relies on the panel or array fabrication process to preserve chemical identity, adding time and expense. As the number of fields in the checkerboard increases, so does the challenge of fabricating the requisite array. In addition, probes must be immobilized—usually by adhesion to the surface of a planar substrate—to maintain the integrity of the spatial encoding scheme. In practice, this assay format can be problematic: sample accumulation can be slow and probe accessibility restricted.
III —Current Applications of Multicolor Fluorescence Detection
The present invention describes a method and apparatus for in-situ interrogation and deconvolution of bead-based combinatorial libraries using multi-color fluorescence imaging and spectral analysis. Recent applications of multi-color fluorescence spectroscopy to DNA sequencing and chromosome painting place requirements on sensitivity and wavelength selectivity exceeding those encountered in conventional applications such as determinations of fluorescence intensity ratios.
Within the context of DNA sequencing, a variety of configurations for rapid detection of 4-color fluorescence have been described. These involve: a dedicated photomultiplier tube detector for each emission wavelength, with corresponding sets of beam splitters in the optical path to produce spatially separated beams; a single detector and rotating filterwheel to select the desired set of wavelengths in a multiplexed recording mode; or a dispersive arrangement that relies on a prism or grating to split the emitted light from multiple fluorophores according to wavelength and takes advantage of recent advances in charge-coupled device (CCD) technology to record spectra on an integrating linear of rectangular CCD array (Karger et al., “Multiwavelength fluorescence detection for DNA sequencing using capillary electrophoresis”, Nucl. Acids Res. 19, 4955 (1991), the contents of which are incorporated herein by reference).